This board is a full ATX, and sports the color-coded PC99-compliant peripheral connectors. The IDE connectors are located near the drive bays, however the floppy connector is at the back of the board behind the processor. The ATX power plug is located at the front of the board, right where we would have expected to see the floppy connector. We found these to be a little awkward, and would have preferred that AOpen swapped the connector locations for slightly easier access but it really isn’t that much of an issue. Since this is a jumperless board, only 3 jumpers are present – Clear CMOS, AGP Ratio and Keyboard/Mouse Wake-up. We didn’t particularly like the location of the Clear CMOS jumper at the back edge of the board under the 3rd PCI card, since this could be a bit of a pain to access. On the other hand, the nice BIOS recovery features help to overcome this issue, as we will discuss later.
The board sports 5 PCI, 2 ISA (1 shared) and 1 AGP slot. All PCI slots are free from obstructions, however the last ISA slot is partially blocked by the front panel connectors, which may prevent a full length card from being installed. 3 DIMM slots limits the maximum memory to 768MB, which may be a consideration for future expansion. There are also headers for SB-Link, Modem Wake-up, Lan Wake-up and IrDA cables.
One other interesting feature, which gives the board part of its name, is the gold heatsink on the North Bridge chip. As more features are included into the motherboard chipset, resulting more transistors being packed in, power dissipation becomes an issue just as it does for processors. Recently, motherboard manufacturers have begun putting heatsinks onto the North Bridge to keep it cool so the system remains stable. AOpen decided to take this one step further and put a gold coating onto the heatsink to help with the heat dissipation. Just as the function of the heatsink is to conduct heataway from the chip, the intent of the gold coating is to conduct the heat from the chipset into the surrounding air more quickly. According to AOpen this will result in a more stable system, particularly when overclocking. This was actually somewhat difficult to prove when you realize that the memory was the limiting factor in our ability to overclock reliably on *any* board using FSB speeds greater than 133MHz.
All CPU settings are made via the BIOS, as is true with most of the newer motherboards. Predefined processor settings are available, as are manual settings for voltage, multiplier and FSB speeds. One nice BIOS feature that AOpen is known for is the ‘Turbo Defaults’ setting, which automatically sets all memory and I/O fields to their highest performing settings. One note of caution – if your components (particularly memory) are not capable of handling these settings, the system could exhibit stability problems.
Both the PHD PCI and PHD Plus diagnostic cards were run in excess of 100 runs. These cards run circuit level diagnostics which test the operation of all the on-board controllers, and the communication between the various components. All tests were successful, with the exception of DMA channel 5, 6 and 7 transfer tests. These exhibited some random failures, most likely due to a small signal timing skew. Since these were intermittent, chances of compatibility problems with some ISA cards are small but still possible. We tested several different cards in the ISA slots without any problem.
We alsoran our QuickTech Pro 2000 software diagnostic for 2 hours without finding any problems. QuickTech Pro is a system level software diagnostic that tests all of the basic system functions (see our review of this software for more information).
Overall, the results of these tests were quite good. The DMA channel errors are not a major issue, due to their intermittent nature. Essentially, the diagnostic card is measuring the signal against a reference timing, and if it is off even a small amount reports a failure. When a particular DMA channel fails on every run, there is more reason to be concerned but an occasionall failure simply means that there is a little more ‘play’ in the signal timings than we would like to see, and it is possible that some ISA cards may not like it.
Operating System Support
Windows 98 (first edition), Windows NT4 (SP4), OS/2 (Warp 3, fix pack 41) and Linux (Red Hat 5.2) were all installed and tested without any problems found. Under Windows 98 we ran Winstone 99 Business Tests for 8 hours without any signs of errors. With Windows NT we ran both Business and High-End Winstone 99 in demo mode for more than 8 hours. Linux and OS/2 were installed without any problems, though no stress testing was performed. We did do a kernel compile in Linux, and ran a small OS/2 benchmark (called SysBench) successfully.
As we would expect from the diagnostics results, the AX6BC Pro Gold is very stable when using standard settings (66MHz or 100MHz FSB speeds). We did run some tests at 133MHz FSB using a non-locked Pentium II 300 (overclocked to run at 3 x 133MHz), and saw no stability problems after 4 hours running Winstone 99 under Windows 98. Bus speeds above 138MHz were essentially unusable because of limitations of the memory.
We found no faster settings than what AOpen already provides by using the ‘Turbo Defaults’ in the BIOS. When we used FSB speeds above 124MHz, we had to back off the SDRAM CAS Latency to prevent data errors from occurring. Above 138MHz, even the standard settings would not allow for reliable memory transfers. Though the BIOS does allow for manual CPU voltage settings, we did not find a need for this.
When using FSB speeds that are non-standard (speeds other than 66MHz and 100MHz), there is a jumper which allows the AGP speed to be set to either full host bus speed, or 2/3 the bus speed. This means that with a 138MHz FSB speed (fastest stable speed using existing memory), the AGP speed will be 92MHz. While it is possible that some AGP cards can handle this, system stability may be a problem and a PCI video card may be advisable for these faster bus speeds. PCI divisors of 2, 3 and 4 are used for the faster speeds so that even at 133MHz FSB the PCI clock is still 33MHz.
AOpen has included some nice recovery features that deserve to be highlighted. The BIOS has a boot block, which allows recovery from a bad flash by using either a PCI or ISA VGA card. Even if the video is not working, it is possible to reflash, as the BIOS will read from the A: drive when using the boot block, and can run an executable. By knowing the sequence of questions and the proper keystrokes it is possible to reflash while ‘driving blind’. Another method would be to create a boot disk and use batch file and the proper switches to automatically answer the questions.
A particularly handy feature for overclockers is the method for resetting the CPU settings to default in the event of a system hang or boot failure after setting the values too ‘aggressively’. The HOME key is held down while the system is reset, and is released when the system begins to POST. We tried this several times, and is a much better method than having to clear the CMOS. As mentioned earlier, this relieves some of the issues with having the clear CMOS jumper in such an inconvenient place, limiting its use to clearing the password (and then we might argue that it *should* require a major effort).
Be the first to discuss this article!